Takehome: Given a dataset and asked to build a model to predict salaries, and write a report on how you solved the problem. It's somewhat open ended, and although they say it should only take ~4 hours, it took longer for me. Video Call: 1 hour video chat with 2 other data scientists. There's a coding question (you have to code live on hackerrank interface and compile), stats/probabilities question, and an experimental design/hypothesis testing question. The coding question is fairly simple (easy to med hackerrank level) involving lists. The stats and probability question was a simple Bayes rule problem, and the the hypothesis testing question was looking at a distribution and answering some questions. On-Site: All day, 4 technical interviews (1 hour each), a data challenge, and a final behavioral interview. First was a stats/probability session with some experimental design (A/B testing) questions, next was white board coding, and a machine learning session. I did fine on those, but then they give you a data challenge with a 90 min time limit. This was the most difficult part for me, as it was about recommendation systems, a topic I was not too familiar with. The final interview is 30min chat with the director/VP guy, and he just asks some behavioral questions.